Design Criteria for Adaptive Roadway Lighting

Introduction

With the development of new lighting technology and a push to reduce the overall energy and environmental impact of lighting, adaptive lighting has become a new trend in the roadway industry. Adaptive lighting is a design methodology in which the light output of a system is adjusted as traffic conditions change. More specifically, roadway lighting illumination levels are adjusted based on the needs of the roadway's users. The level of lighting can be reduced or dimmed when traffic on highways, sidewalks, or both is reduced. Dimming the lighting level while maintaining the lighting configuration will not affect the uniformity of the lighting or an object's contrast; however, contrast thresholds will increase, resulting in longer detection times. In addition, luminaires installed in new lighting designs often exceed the requirements of the lighting design. Over time, the system will meet or exceed the design level even as the lamps age and dirt accumulates on lenses. Adjusting the output of luminaires so the system meets the design level throughout the service life of the lamps can also save on energy costs and eliminate periods of over-lighting. This form of lighting control is not specifically considered adaptive lighting, but an adaptive lighting system provides this option.

The objectives and methodology of this project have been developed to update Reduced Lighting on Freeways During Periods of Low Traffic Density (Publication No. FHWA-RD-86-018), and to develop application guidelines that address the following issues:⁽¹⁾

• Optimal times and conditions for reducing lighting.
• Appropriate lighting levels for various roads and road features.
• Appropriate approaches for reducing lighting.
• Energy savings and reduction in greenhouse gases that may result from reducing lighting.
• Potential legal issues related to reducing lighting, including the development of such issues since the release of the original report.

The resulting design methodology is based on existing Commission Internationale de l'Eclairage (CIE) roadway lighting criteria and provides a process to determine whether adaptive lighting is appropriate for a given roadway.

BACKGROUND

More than 50 percent of all fatal crashes occur at night, even though nighttime volumes are approximately 25 percent of all traffic. The resulting fatality rate for drivers is three times higher at night.^(2,3,4)In general, the introduction of roadway lighting has been found to be a crash countermeasure on highways.

In 1989, Box completed a study comparing the crash rates on a predetermined route (approximately 1.7 mi (2.8 km)) before and after lighting units were installed.⁽⁵⁾The route contained five lanes, with a two-way left turn in the center lane. Several night/day evaluations were conducted to determine the number of crashes occurring at intersections with lighting versus those without lighting. Also of particular interest was the number of crashes occurring in the midblock section. Initial research showed that a total of 146 daytime and 72 nighttime intersection crashes (33 percent of crashes) occurred before the lighting installations were made. A total of 112 daytime and 44 nighttime midblock crashes (28 percent) were also noted. Following the installation of lighting units, 210 daytime and 72 nighttime intersection crashes (26 percent) were reported, and 145 daytime and 31 nighttime midblock crashes (18 percent) were reported. (These numbers include an annualized average daily traffic (AADT) increase of 33 percent between 1984 and 1987.) The ratio of night crashes to day crashes dropped following the lighting installation.

Another study completed by Box examined the relationship between illumination level and freeway crashes.⁽⁶⁾Data were collected on 203 mi (326.7 km) of selected roadways. The selection of the roadways was based on minimum requirements set by the research team: 1) traffic volume data collected for a full year, 2) minimum section length of 1 mi (1.6 km), 3) 1 to 3 years of crash data, and 4) unlit portions of roadway as well as lit portions with complete maintenance and installation records for luminaires. The ratio of night-to-day crashes for lighted and unlighted roadways was 1.43 and 2.37, respectively. Box reported that installing lights on freeways could potentially reduce nighttime crashes by an average of 40 percent and overall crashes by 18 percent. In addition, roadway sections with the lowest illumination range (0.3 to 0.6 horizontal foot candles (HFC)), compared with higher illumination levels (0.8 to 1.1 HFC and 1.3 to 1.5 HFC), had the lowest night-to-day crash ratios. However, Box performed a limited statistical analysis. Figure 1 shows the study's results. Using these results, an analysis of variance statistical test was performed with illuminance as a main effect. The statistical results show that there is no relationship between the number of crashes and the measured illuminance level (F = 0.54, p = 0.663). These results indicate that there is limited applicability of the results from this investigation.

Figure 1 . Graph. Crash ratio versus illuminance (horizontal foot-candles) from Box, 1971.⁽⁵⁾

Removal and reduction of roadway lighting have also been studied. In a study by Hilton in which every third luminaire was turned off, the lighting system still maintained the minimum design illumination (according to roadway standards) despite a 22-percent reduction in illumination. (Note that this reduction of 22 percent from shutting off 33 percent of the luminaires is not expected in terms of the delivered light on the roadway; however, new lamps were installed at the same time.)⁽⁷⁾ In fact, most roadway lighting systems provide more illumination than the design requires. This is because roadway lighting systems are intended to meet the design levels at the end of the lamp service life. Turning off every third luminaire, on a system with new lamps, was partially compensated for by this initial light level. After 2 years, illumination had fallen to 32 percent of the initial value and barely met the 3:1 maximum suggested for uniformity for freeways. (Uniformity Ratio (UR) is the ratio of the average lighting level on the roadway to the minimum lighting level on the road.) Hilton does not provide crash data for the road segment during this evaluation, so it is not known whether the reduction in light level resulted in more crashes. Hilton's findings suggest that standards should consider the impact of depreciation of light over time. Initial light output could be reduced when sources are producing higher light levels than required for the lighting design. As the sources age, the light output can be increased to continue to meet design levels. This would conserve energy over the service life of the lamps. Hilton recommended that illumination levels should be reduced on initialization by one-third for no more than 6 months after activation.⁽⁷⁾

Richards's 1979 case study on Texas highways highlighted the economic consequences of reducing roadway lighting during a time of energy conservation in the early 1970s.⁽⁸⁾A segment of road approximately 7 mi (11.2 km) long on Interstate 35 was chosen for the lighting reduction. Richards found that despite reducing the yearly cost of energy by nearly $25,000, the State estimated an increase in crash costs of $17,000 per year. Richards studied crash rates and costs for 2 years before and 2 years after lighting removal. In the 2 years after lighting removal, Richards cites an increase not only in crash rates but also in crash severity as the rate of fatal crashes increased.⁽⁸⁾ The crash increase led to a crash cost total of $33,880, which exceeded the estimates developed by the State prior to the project.

In 2001, Oregon chose to reduce power consumption by State agencies by 10 percent in response to a perceived energy shortage in the Pacific Northwest. The Oregon Department of Transportation (ODOT) saw the opportunity to conserve energy by selectively reducing illumination on interstate highways within the State. Traffic engineers selected what they perceived to be the safest stretches of highway-based on geometry, crash history, and pedestrian activity-for lighting reduction.⁽⁹⁾

The reduction locations fit into three categories, as listed by Monsere et al.: 1) interchanges where lighting was reduced from full to partial lighting design, 2) interchanges where lighting was reduced from a partial plus design to a partial design, and 3) highway sections where mainline lineal lighting was reduced or removed.⁽⁹⁾ In all, lighting systems at 47 total interchanges and just over 6 mi (9.7 km) of highway were modified. The researchers completed a before-and-after study on the safety effects of these changes. The study includes 9 years of before and 4 years of after data.⁽⁹⁾

When the lighting design was reduced from full to partial, nighttime crashes increased less than 4 percent. When adjusted from full lineal lighting to all or some luminaires turned off, crashes increased nearly 30 percent, and fatal and injury crashes increased nearly 40 percent. When lighting was reduced from partial plus to partial design, crashes decreased by 35 percent; however, Monsere et al. explain that this comparison was the least robust because the sample size was much smaller, and the result should not be interpreted as a definitive safety benefit.⁽⁸⁾Also, the roadways chosen for reductions in lighting designs were selected based on a history of safety, which may have biased the results.⁽⁹⁾

Adaptive Lighting

Control systems are becoming more prevalent for roadway lighting. These systems allow the output of each luminaire to be controlled to avoid over-lighting, which Boyce et al. proposed could initially exceed lighting requirements by as much as 40 percent for newly installed luminaires.⁽¹⁰⁾ (The amount of over-lighting depends on the light source and how much its output declines over time.) These control systems consist of a home base controller and individual modules in each luminaire, which allow for separate and direct control of each luminaire. Development of these types of lighting control systems enables adaptive roadway lighting. Adaptive lighting considers the possible energy conservation when lighting is reduced during non-peak traffic hours. During hours when pedestrian and motor traffic decreases, lighting can, in theory, be reduced without resulting in an increase in crashes or crash severity.

There are various degrees of adaptive lighting. The level of lighting can be reduced or dimmed when traffic on highways, sidewalks, or both is reduced. Reducing the light level will not affect the uniformity of the lighting or an object's contrast; however, the contrast threshold will increase, resulting in a longer detection time. FHWA-RD-86-018, developed in 1985, lists five different methods for reducing lighting to conserve energy.⁽¹⁾ One is to eliminate all lighting after midnight or during other designated periods of reduced pedestrian and motor traffic. Two partial lighting methods involve extinguishing every other luminaire or extinguishing all luminaires on just one side of the roadway. A fourth method is to install two luminaires per light pole and extinguishing one light per pole. The final method, and the one frequently discussed, uses special dimming technology that will reduce the lighting by a predetermined percentage. All five of these methods may be implemented after midnight or another appropriate, designated time. These dimming methods depend on the available technology and the ability to control the lighting system remotely. In addition to adapting light levels based on motor vehicle and pedestrian traffic, luminaire output can be controlled so that the system does not exceed the design levels by adjusting the light output level to respond to lamp aging and dirt depreciation.⁽¹¹⁾

Lighting can be turned off completely if a considerable decrease in traffic rate is expected. However, despite the increased energy savings, this method is not recommended by the CIE.

The lighting can also be adaptive by detecting an increase in roadway users or a need for more lighting and controlling itself accordingly, provided that the technology is available to do so.

For cost efficiency, it is thought that methods involving turning lights completely off, turning off lights on one side, or turning off every other light are the most efficient. These methods reduce energy usage and costs and are more practical for implementation. However, these methods present legal issues associated with whether the resulting lighting meets accepted design criteria after the lighting is reduced and whether changes to the design criteria might affect safety of roadway users.⁽¹⁾

Methods that involve dimming the lights do not conserve as much energy as light-extinguishing methods. However, light-dimming methods are considered neutral regarding their impacts on safety because the resulting light levels meet appropriate design criteria that are believed to meet the requirements of road users under the specified traffic conditions. These methods are considered more prudent because the amount of light is more likely to meet Illuminating Engineering Society of North America (IES) and American National Standards Institute (ANSI) Recommended Practice for Roadway Lighting (ANSI/IES RP-8-00) recommendations.⁽⁴⁾

In 1985, the Federal Highway Administration (FHWA) evaluated detection performance for drivers detecting a small target under different reduced lighting conditions.⁽³⁾The target was a 6- by 6-inch (152.4- by 152.4-mm) square with a smooth, non-reflective surface. The best condition for observing targets from farthest away was full (continuous), unaltered lighting, which had a mean observation distance of 287.9 ft (87.8 m). The mean detection distance for lighting dimmed to 75-percent power was 232.6 ft (70.9 m). When the power was reduced to 50 percent, the mean distance was 223.8 ft (68.2 m). When every other luminaire was extinguished, the distance dropped to 204.8 ft (62.4 m). When only one side of a roadway previously lit on both sides was lit, the mean detection distance dropped to 163.4 ft (49.8 m), a detection distance similar to that when no lighting at all was operational (163.2 ft (49.7 m)). These results indicate that maintaining a proper lighting design, with consistent URs, results in improved target detection distances, which strongly suggests improved safety as compared to extinguishing selected luminaires.⁽³⁾

The cost-benefit analysis observed by the 1985 FHWA study revealed a contrasting result.⁽¹⁾ In terms of annual energy savings per mile, eliminating all roadway lighting was the most conservative and annually saved $1,877 per mi. Extinguishing half of the lights, whether all on one side or every other luminaire, saved approximately $938 per mi. Fixed or variable power dimming up to 50 percent also saved approximately $938 per mi. These results, coupled with the detection results, tend to suggest that reducing the power of the luminaires (dimming) while maintaining uniformity might be the best solution for conserving energy without negatively affecting driver or pedestrian safety.

Current Standards for Adaptive Lighting

The current approach to adaptive lighting is supported by two guidance documents: ANSI/IES RP-8-00 and CIE Document Number 115 (CIE 115).⁽¹¹⁾

The IES method specifies roadway lighting by road type and the potential for pedestrian conflict. The lighting requirements, as of the date of this publication, are shown in table 1. This method adjusts the lighting levels based on changes in the pedestrian conflict level, rather than by reclassifying the roadway.

Table 1 . IES recommended maintained luminance values for roadways.⁽⁴⁾

The CIE method for adapting roadway lighting is much more flexible: roadways are classified as M for motorway, C for conflict areas, and P for roadway with pedestrians.

Using the CIE method to standardize how lighting may be adapted, we must first investigate how non-adaptive lighting levels are derived. Data from the CIE (table 2) detail the parameters for selecting M-class lighting. Values are defined for each parameter option (e.g., very high speed > 65 mi/h (105 km/h)), each parameter value is then given a weighting value (V_W), and the sum of those weighting values (V_WS) is used in the following equation:

Figure 2. Equation. Number of lighting class M.

Table 2 . CIE parameters for selecting M lighting class.⁽¹¹⁾

Table 3 details lighting classes M1 through M6, which cover various roadway conditions and road surface luminance levels. A lighting class is chosen based on summing the weighting values, subtracting from six, and applying the resulting value as the M class. If the resulting number is not a whole number, then the next lowest whole number is used. As an example, if V_WS = 4.5, the lighting class is M1. (6 - 4.5 = 1.5, which is reduced to M1.) If the resulting number is negative, the lighting class is M1.

Table 3 . Lighting classes for motorized traffic.⁽¹¹⁾

In an adaptive lighting implementation, lighting classes change depending on the change in parameters. If traffic volume, intersection density, or ambient luminance fluctuates nightly or even hourly, an adaptive control system can monitor the changes and illuminate accordingly.

It is important to note that other parameters and lighting classes exist; the ones presented here are examples of how lighting classes can be adaptive. CIE 115 contains more classes and specific parameters that pertain to those classes.⁽¹¹⁾

Summary

Previous efforts that investigated the potential for adaptive lighting show that there is a benefit to having the lighting available for both the driver and other road users, but there is the potential to reduce lighting based on established criteria. The next phase of the current effort is to consider the development of specific design criteria for the development of guidelines for adaptive lighting. These design criteria will include both the previous design criteria and methodologies.